Fracture behavior of sisal fiber–reinforced starch-based composites

Fracture Behavior of Sisal Fiber–Reinforced Starch-Based Composites

V. Alvarez, A. Vazquez, C. BernalResearch Institute of Material Science and Technology (INTEMA), Engineering Faculty, National University ofMar del Plata, Juan B. Justo 4302 (B7608FDQ), Mar del Plata, Argentina

The fracture behavior of biodegradable fiber–reinforcedcomposites as a function of fiber content under differentloading conditions was investigated. Composites with dif-ferent fiber content, ranging from 5 to 20 wt%, were pre-pared using commercial starch-based polymer and shortsisal fibers. Quasistatic fracture studies as well as instru-mented falling weight impact tests were performed on thecomposites and the plain matrix. Results showed a signif-icant increase in the crack initiation resistance under qua-sistatic loading. This was caused by the incorporation ofsisal fibers to the matrix and the development of failuremechanisms induced by the presence of the fibers. On theother hand, a modest increasing trend of the resistance tocrack initiation with fiber loading was detected. An im-proved fracture behavior was also observed when the im-pact loading was parallel to the thickness direction. Underthese experimental conditions, the composites exhibitedhigher values of ductility index, energy at initiation and totalfracture energy than the plain matrix. Furthermore, an in-creasing trend of these parameters with fiber content wasdetected in the biocomposites. Overall, the addition ofsisal fibers to the biodegradable matrix appears to be anefficient mean of improving fracture behavior under bothquasistatic and impact loading conditions. POLYM. COM-POS., 26:316–323, 2005. © 2005 Society of Plastics Engineers

INTRODUCTION

Reinforcing thermoplastic polymers with natural fibershas become very attractive because of the good mechanicalproperties that can be obtained at relatively low cost [1].Currently, the use of natural fibers as reinforcements inindustrial applications is specially being considered in theautomotive and packaging industries. The use of naturalfibers as an alternative of the use of glass fibers in biode-gradable matrices is mainly driven by ecological reasons.Several works have been done concerning the use of naturalfibers with thermoplastic biodegradable polymers [2–10].

Natural fibers are commonly used in the form of discon-tinuous reinforcements, although natural fiber–reinforcedpolymers do not offer stiffness and strength values as highas continuous-fiber composites. Their low cost and easyprocessing make them preferred for many applications [11].

On the other hand, short-fiber–reinforced thermoplasticshave a complex microstructure. This is a consequence of thefiber length and orientation distribution in the molding andthe existence of the interface, which affects the mechanicalproperties and the stress concentrations at the fiber ends.

In short-fiber–reinforced composites, a large number ofinelastic failure mechanisms in the vicinity of an advancingcrack become active. These mechanisms contribute to thecreation of the fracture process zone with a concomitantsubstantial slow crack growth preceding unstable fracture[12, 13].

Different energy dissipation mechanisms can be identi-fied depending on the fiber length. In short-fiber–reinforcedthermoplastics, the fibers of subcritical length are pulled outrather than broken, as they are too short to reach theirstrength. In this case, the relevant energy dissipation mech-anisms such as debonding, sliding, restricted pull-out, andbrittle or ductile matrix fracture are well described else-where [11–13]. Other failure mechanisms (i.e., fiber split-ting into ultimate cells, stretching and uncoiling of micro-fibrils in the cells of fibers, transverse microcracking, andmultiple ultimate cell fracture) were also described in nat-ural fiber–reinforced polymers [7, 8].

The fracture characterization of short-fiber–reinforcedpolymers using fracture mechanics has been studied inten-sively [11, 12, 14–18]. However, there are relatively fewreports on the failure behavior of natural fiber–reinforcedbiodegradable polymers [7–10].

In this study, we evaluated the fracture behavior of a sisalfiber–reinforced starch-based polymer as a function of fibercontent under different loading conditions.

METHODOLOGIES FOR FRACTURECHARACTERIZATION

The critical stress intensity factor KIC, and the criticalelastic strain energy release rate GIC, can be determined by

Correspondence to: C. Bernal; e-mail: [email protected] grant sponsor: CONICET-PICT1208011.DOI 10.1002/pc.20103Published online in Wiley InterScience (www.interscience.wiley.com).© 2005 Society of Plastics Engineers

POLYMER COMPOSITES—2005

following the ASTM D 5045-93 recommendations [19].They represent the intensity of the stress field ahead of thecrack tip at fracture initiation and the energy required toinitiate fracture, respectively.

The dependence of fracture toughness on compositioncan be analyzed using the following expressions derived byPukanszky and Maurer [20]. The changes in properties dueto the interaction between components have been consid-ered proportional to the actual values of those properties inthis analysis [20]:

GIc � GIcm�Em/E��1 � vf�/�1 � 2.5vf�exp�BGcvf� (1)

KIc2 � KIcm

2 �1 � vf�/�1 � 2.5vf�exp�BKcvf� (2)

where Em/E accounts for the inverse correlation that existsbetween fracture resistance and stiffness of the material andthe term (1 – vf)/(1 � 2.5vf) represents the change in theeffective load-bearing cross-section of the matrix due to thepresence of the fibers. B is related to fiber-matrix interfaceproperties [21] and is particular to each system.

The effect of the reduced load bearing cross section canbe eliminated, if the reduced properties are plotted againstfiber content in the linearized form of Eqs. 1 and 2:

ln GICred � ln GICm � BGCvf (3)

ln KICred2 � ln KICm

2 � BKCvf (4)

From the slope of the linear regression of experimentaldata, BGc and BKc parameters can be easily determined.Their values can be used for determining the effect ofcomponent interaction on fracture resistance.

Another approach to characterize crack resistance ofmaterials is the work of fracture, which represents theenergy absorbed by the creation of fracture surfaces ofcompletely broken samples. Hence, it involves physicalinformation on the critical load as well as on crack growth[12]. It can be expressed as the total energy supplied to thesample (U) divided by twice the remaining ligament area ofthe test specimen (A) [22]:

wof � U/�2 A� (5)

This test has the advantage that it does not require anyinformation on the stress intensity of the notch, the materialelastic properties, or its mechanical linearity [22].

On the other hand, the material response against impactloads parallel to the thickness direction can be evaluated bythe instrumented falling weight impact technique. Impactstrength and energy at crack initiation as well as totalfracture energy and ductility index can be obtained fromimpact tests [15].

EXPERIMENTAL

MaterBi-Y�, a commercial starch based polymer, waskindly supplied by Novamont, Novara, Italy. Compositeswith different fiber content, ranging from 5 to 20 wt%, wereprepared using MaterBi-Y� as the matrix and short sisalfibers with an average diameter and length of 220 � 30 �mand 7.2 � 0.08 mm, respectively (Brascorda, Brazil). Thepolymer pellets and the sisal fibers were mixed at 180 °C.After mixing for 20 min, the pellets were compression-molded into plaques at 180 °C and 700 MPa and rapidlycooled down with running water. Finally, the plaques wereannealed in an oven for 1 h at 60°C in order to releasethermal stresses generated during molding.

It is important to note that fiber fragmentation can beinduced during the above procedure. Average diameter andlength values of sisal fiber after processing have been re-ported previously [23] (fiber diameter � 130 � 50 �m andfiber length � 2.42 � 0.98 mm). They were measured fromoptical microscopy after extracting the fibers from matrixwith acetone.

Quasistatic fracture characterization was performed onthree-point bend specimens (SE(B)) cut out from compres-sion-molded thick plaques (thickness, B � 5 mm) whichwere machined to reach final dimensions and improve edgesurface finishing.

Sharp notches were introduced by sliding a fresh razorblade into a machined slot. Crack-to-depth ratio (a/W) was0.5. Thickness-to-depth ratio (B/W) and span-to-depth ratio(S/W) were always kept equal to 0.5 and 4, respectively.

Fracture tests were performed in an INSTRON dyna-mometer 4467 with a crosshead speed of 2 mm/min.

Puncture tests were conducted on disk samples of 90-mmdiameter cut out from 3-mm thickness compression-moldedplates. These tests were performed in a falling weight Frac-tovis of Ceast at 1 m/s.

All samples were dried in a vacuum oven at 60°C untilconstant weight was attained before testing. All tests werecarried out at room temperature.

Fracture surfaces of SE(B) samples were characterizedby scanning electron microscopy (SEM).

RESULTS AND DISCUSSION

Quasistatic In-Plane Fracture Properties

Figure 1 shows load-displacement traces of theMaterBi-Y� matrix and the composite with 15 wt% of fiberstested in three-point bend configuration. The MaterBi-Y�matrix displayed almost linear load-displacement traceswith an abrupt drop of load to zero immediately after crackinitiation, revealing the brittle-like behavior of this material.Conversely, the sisal reinforced composites showed higherstiffness and a higher degree of nonlinearity before maxi-mum load caused by irreversible events [24]. A gradual fallof load from the maximum point was observed thereafter,and samples exhibited stable crack propagation. Further-

POLYMER COMPOSITES—2005 317

more, no signs of macroscopic yielding of the matrix wereobserved on tested samples. Linear elastic fracture mechan-ics (LEFM) through the stress intensity factor (KIQ) and thestrain energy release rate (GIQ) was adopted to characterizethe resistance to crack initiation in these materials. Theseparameters were calculated from the maximum in the load-displacement curves following the procedure described inthe ASTM standard [19]. The results are listed in Table 1along with their deviations. A significant increase in KIQ

and GIQ was observed as a result of the addition of sisalfibers to the plain matrix and the development of new failuremechanisms induced by the presence of those fibers. Amodest increasing trend of these fracture parameters withfiber loading was also found as a consequence of the in-crease of these dissipation mechanisms.

The fracture parameter values obtained here did notrepresent valid plane strain fracture toughness values, sincethe materials load-displacement behavior was near the limitof validity of LEFM (10% nonlinearity allowance) and theASTM size requirement for plane strain determinations [19]was not fulfilled (Table 1). However, our fracture parametervalues still reflected a critical stress state for crack initiation[25]. (The minimum thickness values Bmin in Table 1 werecalculated using tensile strength values reported in Ref. 26).

The work of fracture approach, which mainly describesthe stage of crack propagation [12], was also performed in

order to detect any difference in the fracture propagationmode that was not reflected by the fracture initiation pa-rameters [27]. The higher values of the work of fractureexhibited by the composites compared to the unreinforcedmatrix (Table 1) indicated that the incorporation of sisalfibers to the matrix also improved fracture propagationbehavior in our composites. On the other hand, the work offracture appeared to be independent of sisal fiber contentsuggesting that no significant differences in the fracturepropagation mode existed among these biocomposites.

Figure 2 shows the results obtained from the applicationof Pukanszky and Maurer model [20] for KIQ and GIQ onour sisal/MaterBi-Y� composites. A substantial linear cor-relation was obtained for both fracture parameters suggest-ing that this model successfully fit our experimental data.

FIG. 1. Load-displacement traces for the MaterBi-Y� matrix and thecomposite with 10 wt% of sisal fibers.

TABLE 1. In plane fracture properties obtained under quasi-staticloading conditions.

Sisal fiberswt% GIQ (kJ/m2)

KIQ

(MPa � m1/2)Bmin

(mm) wof (KJ/m2)

0 0.78 � 0.12 0.86 � 0.11 11.6 0.77 � 0.115 1.53 � 0.09 1.43 � 0.18 24.6 1.48 � 0.21

10 1.59 � 0.11 1.70 � 0.14 29.3 1.49 � 0.1415 1.61 � 0.12 1.97 � 0.18 34.4 1.53 � 0.0720 1.68 � 0.16 2.04 � 0.16 30.7 1.48 � 0.06

FIG. 2. a: Reduced values of KIQ (MPam1/2) plotted in a linearized formaccording to Eq. 3. b: Reduced values of GIQ (KJ/m2) plotted in a linear-ized form according to Eq. 4.

318 POLYMER COMPOSITES—2005

Therefore, the corresponding parameters BKc and BGc couldalso be used to study the effect of interaction on fractureresistance in these composites. Based on these results, thePukanszky and Maurer model [20] was also used to fit theimpact fracture data previously reported in the literature forthe same composites containing treated and nontreated fi-bers [26]. A substantial linear regression of these data isshown in Fig. 3, (r was 0.9776 and 0.9590 for the compos-ites with nontreated and treated fibers, respectively). Thus,it also confirms the assumption that the Pukanszky andMaurer model [20] can adequately describe the compositiondependence of fracture parameters in our composites. How-ever, the difference in the interaction parameter values BGC

was within the experimental scatter of impact data, and wasunable to quantify the small improvement of fiber-matrixinterface properties obtained with alkali treatment of sisalfibers [26].

Impact Out of Plane Fracture Properties

Table 2 shows the values of disk strength, energy atcrack initiation, ductility index, and fracture energy of thedifferent composites assayed. A decrease in the disk

strength in comparison to the plain matrix was observed forthe composite containing up to 15 wt% of sisal fibers. Wehypothesize that the fibers spread in the matrix can act ascrack initiation points during impact for these compositions[15]. In addition, a regular increasing trend of disk strengthwith fiber content was observed. Indeed, there were enoughfibers at 20 wt% of sisal to enhance load transfer [15].Conversely, the addition of sisal fibers to the MaterBi-Y�matrix led to a sharp increase in the ductility index (Table 2)as new energy dissipation mechanisms resulting from thepresence of the fibers became active. The same effect wasobserved by Mouzakis et al. [15] with injection moldedPP/DLGF composites. For the composites investigated here,the ductility index was found to increase with fiber loadingspecially for the highest fiber content used (20 wt%).

A similar trend for the energy at initiation and the frac-ture energy with fiber content was also observed. This resultsuggests that the incorporation of sisal fibers to the biode-gradable matrix is also an efficient way to improve impactfracture properties.

Figure 4 shows the macrophotographs of the oppositeside of the impacted areas of the neat polymer and thecomposites with 10 and 15 wt% of sisal fibers. Many radialcracks were observed around the hole area of the matrixsamples (Fig. 4a), whereas in the composites a circumfer-ential cracking became more dominant as fiber contentincreased (Fig. 4b and c). Mouzakis et al. [15] have pointedout that the circumferential shear-cracking phenomenon ismore effective in toughness improvement than radial crack-ing mechanism. Hence, the increasing trend of fractureenergy with fiber content shown in Table 2 could be ex-plained in terms of an increase in the circumferential shear-cracking mechanism as the fiber content increased.

Fracture Surface Analysis

The SEM micrographs of fracture surfaces of SE(B)samples of the composite with 15 wt% of sisal are shown inFig. 5a–d at different magnifications. In all micrographs,the matrix displayed flat surfaces indicating the absence ofany plastic deformation. Furthermore, sisal fibers did nothold the matrix material, suggesting a relatively poor fiber-matrix interface.

The presence of fiber pull-out could be predicted byconsidering the critical fiber length concept. For short fiber

FIG. 3. Reduced values of GIQ (KJ/m2) plotted in a linearized formaccording to Eq. 4 for the impact data reported in literature for the samecomposites [26].

TABLE 2. Impact out of plane fracture properties.

Sisal fibers wt%Disk strength,

� (MPa)Energy at initiation,

Einit (kJ/m)Ductility index,

D.I.Fracture energy,

Etotal (kJ/m)

0 111.0 � 5.44 0.17 � 0.03 0 0.17 � 0.035 80.1 � 1.43 0.33 � 0.09 0.14 � 0.04 0.40 � 0.13

10 97.1 � 1.36 0.37 � 0.03 0.16 � 0.03 0.43 � 0.4415 106.4 � 1.85 0.37 � 0.04 0.18 � 0.05 0.45 � 0.0320 129.8 � 7.59 0.46 � 0.02 0.24 � 0.01 0.61 � 0.03


reinforced composites, there is effective load transfer fromthe matrix to the fiber provided a strong fiber-matrix inter-facial bond and a fiber length longer than a critical value(Lc) exist [28]. Lc can be estimated by applying the Kelly-Tyson model [29]:

Lc � d�f/ 2� (6)

where �f is the fiber strength, d is the fiber diameter, and �is the interfacial shear strength. For the sisal fiber used forpreparing our composites, �f � 471.6 MPa [26] and forperfectly bonded fibers, � was assumed to be the shearstrength of the matrix, which was 6.3 MPa for MaterBi-Y�[23]. Lc was found to be about 4.9 mm, which was longerthan the mean fiber length of the composites after process-ing (L � 2.42 mm). Hence, the fibers were expected to bepulled out rather than broken as they were not able to reachtheir strength.

Figure 5 shows fiber pull-out (Fig. 5a and b) as well asother energy dissipation mechanisms which are active andproduce the increase in toughening. Such mechanisms are:modest fiber-matrix debonding, separation of sisal fibersinto ultimate cells by axial splitting of the boundary layer,uncoiling of these spirally arranged microfibrils (Fig. 5c),and microcracking of ultimate cells (Fig. 5d). It has beenpreviously reported by Lu et al. [8] that even uncoiling ofmicrofibrils inside the plant fibers consumes substantialenergy and therefore, imparts high mechanical performanceto the composite. On the other hand, surface roughness isalso clearly observed in Fig. 5. This is due to the organicmatrix surrounding the primary cell wall of fibers, whichmight be responsible of additional energy absorbing mech-anisms such us friction and pull-out of ultimate cells [30].

CONCLUSIONS

The fracture behavior of completely biodegradable com-posites was studied as a function of fiber content underdifferent loading conditions.

Under quasistatic loading, the biocomposites exhibitedhigher resistance to crack initiation and work of fracture incomparison to the plain matrix. These differences betweenthe composites and the matrix may be due to the develop-ment of new energy dissipation mechanisms derived fromthe presence of the fibers. In addition, the increase of thesefailure mechanisms may explain the slight increasing trendof the resistance to crack initiation with fiber loading.

The simple model proposed by Pukanszky and Maurer[20] to predict fracture properties in heterogeneous systemswas suitable for these sisal/MaterBi-Y� composites.

We have also found that the impact out of plane fractureshowed an increasing trend of energy at initiation, ductilityindex, and total fracture energy proportional to the fiberloading. The total fracture energy response was interpreted

FIG. 4. Macrophotographs of the opposite of impacted sides. a:MaterBi-Y� matrix. b: Composite with 10 wt% of fibers. c: Compositewith 15 wt% of fibers.


in terms of the increasing circumferential shear-crackingmechanism observed on impacted samples.

The fracture surface analysis suggested that axial split-ting of the boundary layer between ultimate cells, uncoilingof microfibrils inside the plant fibers, microcracking, and

fiber pull-out were the main failure mechanisms active inthese composites.

In conclusion, the results obtained in this study showedthat the addition of sisal fibers to the actual biodegradablematrix appears to be an efficient mean of improving fracture

FIG. 5. SEM micrographs of fracture surfaces of SE(B) samples of the composite with 15 wt% of fibers atdifferent magnifications. a: 20�; b: 48�; c: 100�; d: 500�.


behavior under both quasistatic and impact loading condi-tions.

REFERENCES

1. A.K. Bledzki and J. Gassan, Prog. Polym. Sci., 24, 221 (l999).

2. S. Iannace, L. Nocilla, and L. Nicolais, J. Appl. Polym. Sci.,73, 585 (1999).

3. H. Hanselka, and A.S. Herrman, “Fiber Properties and Char-acterization,” 7 Internationales Techtexil Symposium, 20(1994).

4. A. Dufresne and M.R.Vignon, Macromolecules, 31, 2693(1998).

5. J. Kuruvilla and L.H.C.Mattoso, “Sisal Fibre Reinforced Poly-mer Composites: Status and Future,” Third Int. SymposiumNatural Polymers and Composites, 333 (2000).

6. C. Biastioli, Principles and Applications: Starch-PolymerComposites in Degradable Polymers, G. Scott and D. Gilead,editors, Chapman & Hall, London (1995).

7. G. Romhany, J. Karger-Kocsis, and T. Czigany, Macromol.Mater. Eng., 288, 699 (2003).

8. X. Lu, M.Q. Zhang, M.Z. Rong, G. Shi, and G.C. Yang,Polym. Compos., 23, 624 (2002).

FIG. 5. (Continued)


9. X. Lu, M.Q. Zhang, M.Z. Rong, G. Shi, G.C. Yang, and H.M.Zeng, Adv. Compos. Lett., 8, 231 (1999).

10. M. Avella, E. Martuscelli, B. Pascucci, M. Raimo, B. Focher,and A. Marzetti, J. Appl. Polym. Sci., 49, 2091 (1993).

11. B. Lauke, J. Polym. Eng., 11, 104 (1992).

12. B. Lauke and W. Pompe, Compos. Sci. Technol., 26, 37(1986).

13. B. Lauke and W. Pompe, Compos. Sci. Technol., 31, 25(1988).

14. S.-C. Wong and Y.-W. Mai, Polym. Eng. Sci., 39, 356 (1999).

15. D.E. Mouzakis, T. Harmia, and J. Karger-Kocsis, Polym.Polym. Comp., 8, 167 (2000).

16. S.-C. Wong, G.-X. Sui, C.-Y. Yue, and Y.-W. Mai, J. Mat.Sci., 37, 2659 (2002).

17. G.-X. Sui, S.-C. Wong, and C.-Y. Yue, Compos. Sci. Technol.,61, 2481 (2001).

18. S.-C. Tjong, S.-A. Xu, R.K.-Y. Li, and Y.-W. Mai, Compos.Sci. Technol., 62, 831 (2002).

19. ASTM D5045-93, Standard Test Methods for Plane-StrainFracture Toughness and Energy Release Rate Determinationof Plastics Materials, American Society for Testing and Ma-terials, Philadelphia (1993).

20. B. Pukanszky and F.H. Maurer, Polymer, 36, 1617 (1995).

21. A. Bezeredi, Z. Demjen, and B. Pukanszky, Die AngewandteMakrom. Chemie, 256, 61 (1998).

22. M. Sakai and H. Ichikawa, Int. J. Fract., 55, 65 (1992).

23. V. Alvarez, A. Terenzi, J.M. Kenny, and A. Vazquez, Polym.Eng. Sci., 44, 1907 (2004).

24. M. Hughes, C.A.S. Hill, and J.R.B. Hague, J. Mater. Sci., 37,4669 (2002).

25. R. Gensler, C.J.G. Plummer, C. Grein, and H.H. Kausch,Polymer, 41, 3809 (2000).

26. V. Alvarez, R. Ruseckaite, and A. Vazquez, J. Comp. Mater.,37, 1575 (2003).

27. P. Montemartini, T. Cuadrado, and P. Frontini, J. Mater. Sci.,10, 309 (1999).

28. S.C. Tjong, S.A. Xu, and Y.-W. Mai, Mater. Sci. Eng., A347,338 (2003).

29. S.G. Wong, G.X. Sui, Y. Yue, and Y.-W. Mai, J. Mater. Sci.,37, 2659 (2002).

30. A. Stamboulis, C. Baillie, and E. Schulz, Die AngewandteMakrom. Chemie, 272, 117 (1999).


Documents

Fracture behavior of sisal fiber–reinforced starch-based composites